001-40 WC 143 JULY-AUG25 PT - Flipbook - Page 16
CONSERVATION
“Thisisnotjustatoolfortoday’sproblems.It’sbuilding
theevidencebasewe’llneedforthenextdecade.”
This satellite image, taken on August 3, 2024, shows the Grand River 昀氀owing into Lake Erie. While the river appears mostly clear, an anomaly
at its mouth suggests potential pollution sources, including marine activity in Feeder Canal, e昀툀uents from Innophos, and contaminants from
a marine warehouse.
“The traditional approach is like checking the temperature of a sick patient once a month,” says Varedi.
“You miss everything in between. If you want to treat
the problem effectively, you need real-time insight—and
that’s what we provide.”
Central to the Waterlix solution is its ability to process and analyze massive amounts of raw satellite data.
Instead of pre-possessed satellite imagery, Waterlix starts
from the original sensor data. This allows it to correct for
atmospheric interference like haze, humidity, or cloud
cover, and to detect small-scale anomalies that might
otherwise be missed.
These anomalies are then passed through machine
learning models trained to identify spectral signatures
of pollutants such as algal blooms, chemical discharges,
heavy metals, or organic contaminants. In many cases,
the system can differentiate between natural variations
and human-caused pollution, offering a level of specificity that greatly improves decision-making.
The system also investigates likely contributors using
upstream geospatial analysis, permitting databases, and
16
WATER C AN ADA • JULY/AUGUS T 2025
documented pollution history. In one deployment in
Brazil, spectral data from a river suggested the presence
of solvents. By cross-referencing with regional industrial
activity and previous discharge records, the team was
able to isolate a cluster of facilities as the probable source
within minutes of the satellite pass or information from
real-time sensors.
Waterlix has supported water quality monitoring in
diverse and high-impact systems: the sacred Ganges and
Kaveri Rivers in India, the Seine in France, Rio Paraíba
do Sul in Brazil, Lake Tahoe in the U.S., and Canada’s
own Lake Erie and the St. Lawrence River. In each case,
the platform provided both local specificity and regional
context—key ingredients in designing meaningful responses.
In Lake Erie, for example, the analysis challenged
prevailing assumptions that agricultural runoff was the
main cause of water quality degradation. Instead, the
data revealed a more complex picture in which industrial
and urban discharges were playing an underreported but
significant role.
WAT E R C A N A D A . N E T
